Search *

[margaretspring]

Summary

Grant seekers can search, filter, and sort opportunities on simpler.grants.gov

Press release

We are pleased to announce that Simpler.Grants.gov now has improved search capabilities that will make it easier for the public to find grant opportunities published by Grants.gov. These features include more relevant search results based on keywords, filtering and sorting, the ability to easily share and bookmark search results, and an easier to use interface.

Test users of the new search page reported that they find it to be a significant improvement over existing grants search options.

Acceptance criteria

• [ ] Search, filter, and sort: Users can
  • [x] search opportunities by keyword
  • [x] filter by at least 3 structured fields
  • [ ] sort by relevance and 3 additional fields
• [x] Stateful UI: Search criteria are reflected in the URL, so that users can bookmark or share the link to a page with a given combination of search criteria.
• [ ] API parity: All search features available via the UI are also available via API.
• [ ] Load testing: Load testing performed to assess whether we can simulate 2,000 concurrent users with our current infrastructure.

Metrics

• [ ] Adoption: At least 10,000 unique users have used the search UI, and at least 5 unique users have used the search directly from the API endpoint (i.e., not via the UI).
• [ ] Reliability: The monthly average error rate remains below 5%, meaning fewer than five 5xx errors per 100 API and five 5xx errors per 100 page requests in production.
• [ ] Search accuracy: Set a baseline metric for accuracy, as measured by monthly click-through rate (CTR) for results on the first page.

Assumptions

Dependencies

[lucasmbrown-usds]

FYI, I updated the Search accuracy metric to reflect our proposed new language about setting a baseline.

[mxk0]

Internal notes:

• What dependencies does making the search API available introduce? Do we need basic API auth?
• Likely use static API keys for a hand-picked set of API users for Quad 1

[widal001]

Internal notes:

• What dependencies does making the search API available introduce? Do we need basic API auth?
• Likely use static API keys for a hand-picked set of API users for Quad 1

Great callout! Maybe we can touch on this in the "Assumptions" section or in the "Technical description" section of the agile estimates?

[mxk0]

@widal001 we made #2244 to cover those questions, so I think we're good!

[mxk0]

@widal001 re: the new "Load testing" acceptance criteria, I don't think 10,000 concurrent users is the goal. To hit this deliverable we need 10,000 unique users for search, but that's cumulative rather than concurrent. We have a load testing spike (#2322) where we'll think through what peak concurrent usage could look like in order to get our 10,000 cumulative users. Can we cut the load testing criterion? We don't know what potential peak concurrent usage is yet, so I don't think it makes sense to set a number as an acceptance criterion.

Same comment goes for #2203. The 2,000 user goal is cumulative, not concurrent.

cc @mdragon

[margaretspring]

@widal001 - I'd like to get the Deliverable approval over to Julius. Can you clarify these item @mxk0 asked about re: the addition load item that wasn't in the HHS Priorities presentation. Thanks!

[mxk0]

For the record, I booked time for us to discuss the open questions tomorrow.

[mxk0]

Sprint 1.2 updates

Status: 🟢 On track

Sprint goal

The main goal for this deliverable in Sprint 1.2 was to deploy OpenSearch.

Accomplishments

Search launch

• Deployed OpenSearch to all environments
• Ran ELT process to populate all environments with data
• Defined launch plan and scheduled syncs to hit our launch milestones
• Designed sort by relevancy
• Shipped UI responsive layout improvements
• Switched frontend to use v1 Search endpoint
• Created load testing plan

Generate traffic for search

• Developed content for email outreach and got feedback from stakeholders
• Submitted comms requests to get emails sent

Rollover

• Implementation for sort by relevancy

Risks

• We haven't worked with comms for email outreach before. This process is going well so far, but we may run into hurdles.

Sprint 1.3 goals

• Publicly launch Search!